The different kinds of confidence

Hello! This is my first ever LessWrong post!

The following is extracted from a conversation I had with DaystarEld on Facebook, and I have his permission to share it on LessWrong.

tl;dr: Personal skills you can be confident about can be broken down into three types:

Absolute skill: Are you good at what you do, relative to the general population?
Relative skill: Are you good at what you do, relative to other people who do the same thing?
Instrumental skill: Are your skills applicable to this situation? Is it part of the reference class, or do you need to learn/try something else? Should you ask someone else, or try and handle it yourself?

DaystarEld asserted that personal confidence is not just about your own belief in your abilities, but also your perceptionn of common knowledge of your capabilities.

DaystarEld: I’ve had experiences where I was confident in my own competence, but was not sure those I would be working with were aware of it, or whether they would value the things I was competent in, and this effectively served to make me hesitant and reserved until I received enough positive signals from those around me, and then felt more comfortable acting with agency and status.

The more I think about it, the more I wonder about how generalized this is. Because clearly there are people who act confident but do not have their competencies acknowledged by those around them, or just have no reason to believe that it is yet.

He wondered if someone who is well-calibrated as to their ability would be considered overconfident or arrogant if they acted confidently without social signals of respect.

DaystarEld: I’m curious to know if there another word for being overconfident in how your abilities are perceived by others, regardless of how accurate your own perceptions of your abilities are. And is it a GOOD thing to not act confident when you lack social signals of common knowledge? Or should people care more about their own calibration of their abilities than what others think? I can see cases for both, but on a purely what-gets-best-results level, how valuable have others found it to ramp up your confidence with each individual or group of people you interact with, opposed to starting at the point that you feel you most deserve?

I reasoned that, assuming I’m well-calibrated and confident in my ability, others not acknowledging your skill shouldn’t really make a difference in your confidence unless they give you evidence against it, or their opinions should be weighted more heavily because e.g., they are more exprienced in the field.

If I’m confident in an ability, why would I care about whether others acknowledge it? Or, instead, why should others acknowledging it or not have any effect on my confidence?

On the other hand, if they give me evidence against it (e.g., I consistently make mistakes, or my reasoning/understanding is wrong), then my confidence in my ability should drop.

is it a GOOD thing to not act confident when you lack social signals of common knowledge?

I’m not sure it’s common at all. I’d like it if you could give me a few examples of where people would act like that.

That being said, it might be context dependent. If your belief goes against established (scientific) consensus, then you wouldn’t get much support (though you might get some among untrained people). This should indicate that your position is wrong. It’s basically a lack of calibration.

Another thing is the Dunning-Kruger effect. You can be confident and not skilled. That is a miscalibration again.

I think that, until you have a track record of accuracy in your perception of your abilities, your priors are weak enough that perception by others (or lack of buy-in) (weighted by how likely they are to be correct) should have a big effect on your confidence. After you have strong priors, though, it needs a much bigger input to move your confidence as much.

To give a real life example, if you’re a therapist who’s consistently delivering good results, or are able to help people at your work, but your patients often start from a position that the whole thing is a scam, would you reset to “I’m not that good” with each new person?

Here, DaystarEld replied that he was wondering if the core of condience itself is one that’s dependent on epistemology, since that doesn’t appear to be the case for most people. He also gave some examples of where he thought his concerns might apply.

DaystarEld: To be a bit more clear, I think this is a separate thing than the idea of how confident one should be about their own abilities. I agree that there’s overlap in that signals from others should be important in determining how justified self-confidence is, but I’m curious about whether the core of confidence itself is one that’s dependent on epistemology. For most people it doesn’t appear to be.

>I’m not sure it’s common at all. I’d like it if you could give me a few examples of where people would act like that.

I actually don’t know how common it is, but it seems like the kind of thing that you might see when someone goes from the best-among-people-they-know to a situation where they’re more among potential equals? Or maybe someone who just changes context of where/how they do a thing, so if a singer from one country who is decently popular goes to another country where they’re not sure if the songs they sing are appreciated or liked? I have other examples that all seem really personal to me, which is partly why I’m not sure how general this is.

>To give a real life example, if you’re consistently delivering good results, or are able to help people at your work, but your patients usually start from a position that the whole thing is a scam, would you reset to “I’m not that good” with each new patient?

It’s less “I’m not that good,” and more an attitude of being more cautious and less assertive in the things I would say or suggest. I think this is in part because of the natural rapport that needs to be re-established each time, but also in part a check against whether whatever bundle of skills and knowledge I have for therapy is actually recognized as applicable or useful to each individual. I think.

Here, I thought that he might be combining various concepts under the word “confidence,” and suggested that it be broken into three.

In the first case (e.g., going to a top university), it’s just miscalibration, since while you’re best-in-class in high school, the class of problems you’ll face is different, and you can’t necessarily expect to be good at solving these. That being said, even the worst player in the World Cup is probably a much better football player than someone from the general population, so they should have confidence in their absolute skill, if not their relative skill.

In the second case (i.e., the singer performing in a new country), there tends to be market research done on that (you wouldn’t normally send the singer and the equipment and all the support staff and their equipment to another country on a whim). Even indies could at least put a song on Youtube and see where the viewers are coming from.

They can be sure that they know how to sing well. They can be sure that they know they are good with their home audience, but they don’t know yet whether they will be good with their new one. And that’s fine. If they want to get that audience, they might need to practice a new style or come up with something different. Or, they might decide not to invest their time and energy into this particular subset of the population.

I think, in this context, it’s important not to conflate absolute skill, relative skill, and instrumental skill.

It’s less “I’m not that good,” and more an attitude of being more cautious and less assertive in the things I would say or suggest. I think this is in part because of the natural rapport that needs to be re-established each time, but also in part a check against whether whatever bundle of skills and knowledge I have for therapy is actually recognized as applicable or useful to each individual. I think.

And that’s great! Let’s break it down into the above skills:

Absolute skill: Are you good at your job, relative to the general population? Yes.
Relative skill: Are you good at your job, relative to other people in your profession? You know better than I do. I won’t ask you to comment publicly.
Instrumental skill: Are your skills applicable to this patient? Are they part of the reference class, or do you need to learn/try something else? Should you refer them to someone else, or try and handle it yourself?

And don’t forget that the fact that you notice that and try to work around it does require a certain base amount of skill, and it’s more likely to lead to success both for this patient, and similar ones that come in the future.

I’m curious about whether the core of confidence itself is one that’s dependent on epistemology

For me, at least, I think it is. I don’t see where public perception comes into play. If you’re good at your job, you know what you’re good at, and where you’re weak at. In other words, you’re confident in your assessment of your skills. The applicability of your skill to the situation, and your ability to compensate if it’s not necessarily applicable, are things you should be able to determine with a different level of confidence.

Does this make sense?

In the end, DaystarEld liked the destinction, but felt that there may still be some element of confidence that has to do with social awareness, status, or signaling, even if it might not be optimal or applicable for all people or situations. That being said, I’m notoriously bad at even noticing social signals in the first place, so it could be that I just missed it naturally.

What do you think?

edit: Temporarily added a link to the Guild of the ROSE.